Skip to content
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension


Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
8 changes: 4 additions & 4 deletions CHANGELOG.md
Original file line number Diff line number Diff line change
Expand Up @@ -735,15 +735,15 @@ hunting for the wrong things.

Plan §1 goal-4 ("no regression on CLI hot path vs the v0.5.35
baseline") verified end-to-end on the Windows 7-drive reference
box. Current v0.5.100 (post-Phase-8 tiered architecture) is
box. Current v0.5.102 (post-Phase-8 tiered architecture) is
**universally faster** than v0.5.35 across every benchmarked
pattern, with the largest result set (`*.dll`, 44 529 rows)
showing a **2.7× speedup**:

```
Drive D, 7.07 M records, 30 rounds, HOT phase, p50 / p95 wall_ms:

v0.5.35 v0.5.100 Δ p50
v0.5.35 v0.5.102 Δ p50
exact (3 rows) 20 / 23 → 18 / 19 −10 %
prefix (8 732) 46 / 50 → 40 / 46 −13 %
ext_rare (11) 18 / 20 → 17 / 18 −6 %
Expand Down Expand Up @@ -919,7 +919,7 @@ log-message renames fail CI before reaching another 24-h soak.
2026-05-13. No new operator-surface features land on `main`
until v0.6.0 ships.

## [0.5.100] - 2026-05-08
## [0.5.102] - 2026-05-08

> **Note on the v0.5.91 gap.** v0.5.91 was prepared and tagged but never
> reached a published GitHub Release: the `release.yml` finalize step hit
Expand All @@ -928,7 +928,7 @@ log-message renames fail CI before reaching another 24-h soak.
> partial release was deleted, the tag name became permanently locked by
> GitHub's *immutable releases* feature (the pre-receive hook refuses any
> future ref creation under that name even after a clean delete). The
> public release sequence therefore jumps `v0.5.90 → v0.5.100`; all
> public release sequence therefore jumps `v0.5.90 → v0.5.102`; all
> intended v0.5.91 changes are rolled forward into this release.

### Fixed
Expand Down
Loading
Loading